def convert_to_csv(self, remove=True):
"""
This function converts the tab-delimited txt file from Dragon to pandas dataframe.
Note that this process might require large memory based on the number of data points and features.
Parameters
----------
remove : bool, optional (default = True)
if True, the original descriptors file (Dragon_descriptors.txt) will be removed.
Returns
-------
features : pandas DataFrame
The 2D dataframe of the descriptors. Note that the first two columns are 'No.' and 'NAME'.
"""
# convert to csv file
t0 = time.time()
print("converting output file to csv format ...")
df = pd.read_csv(self.data_path, sep=None, engine='python')
# df = df.drop(['No.', 'NAME'], axis=1)
# execution time
tmp_str = tot_exec_time_str(t0)
# remove original tab delimited file
if remove:
os.remove(self.data_path)
self.data_path = None
print("... conversion completed in %s" % tmp_str)
return df
def run(self):
t0 = time.time()
print("running Dragon%i ..." % self.version)
os_ret = os.system(
'nohup dragon%sshell -s %s' %
(self.version, os.path.join(self.output_directory, self.drs_name)))
if os_ret != 0:
msg = "Oops, dragon%ishell command didn't work! Are you sure Dragon%i software is installed on your machine?" % (
self.version, self.version)
raise ImportError(msg)
# execution time
tmp_str = tot_exec_time_str(t0)
print("... Dragon job completed in %s" % tmp_str)
def test_exec_time_str():
time_start = time.time()
time.sleep(0.5)
s = tot_exec_time_str(time_start)
assert int(s[-4]) >= 5
def call(self):
self.refs = {}
for iblock in self.ImpOrder:
task = self.cmls[iblock]['task']
parameters = self.cmls[iblock]['parameters']
host = parameters.pop('host')
function = parameters.pop('function')
start_time = time.time()
tmp_str = "======= block#%i: (%s, %s)" % (iblock + 1, host,
function)
print(tmp_str)
tmp_str = "| run ...\n"
print(tmp_str)
if host == 'sklearn':
# check methods
legal_functions = [
klass[0] for klass in inspect.getmembers(sklw)
]
if function in legal_functions:
self.references(host, function) # check references
self.Base.graph_info[iblock] = (host, function)
cml_interface = [
klass[1] for klass in inspect.getmembers(sklw)
if klass[0] == function
][0]
cmli = cml_interface(self.Base, parameters, iblock, task,
function, host)
cmli.run()
else:
self.references(host, function) # check references
self.Base.graph_info[iblock] = (host, function)
cml_interface = [
klass[1] for klass in inspect.getmembers(sklw)
if klass[0] == 'automatic_run'
][0]
cmli = cml_interface(self.Base, parameters, iblock, task,
function, host)
cmli.run()
elif host == 'chemml':
# check methods
legal_functions = [
klass[0] for klass in inspect.getmembers(cmlw)
]
# print("legal_functions: ", legal_functions)
if function not in legal_functions:
msg = "@function #%i: couldn't find function '%s' in the module '%s' wrapper" % (
iblock, function, host)
raise NameError(msg)
self.references(host, function) # check references
self.Base.graph_info[iblock] = (host, function)
cml_interface = [
klass[1] for klass in inspect.getmembers(cmlw)
if klass[0] == function
][0]
cmli = cml_interface(self.Base, parameters, iblock, task,
function, host)
cmli.run()
elif host == 'pandas':
# check methods
legal_functions = [
klass[0] for klass in inspect.getmembers(pdw)
]
if function not in legal_functions:
msg = "@function #%i: couldn't find function '%s' in the module '%s' wrarpper" % (
iblock, function, host)
raise NameError(msg)
self.references(host, function) # check references
self.Base.graph_info[iblock] = (host, function)
cml_interface = [
klass[1] for klass in inspect.getmembers(pdw)
if klass[0] == function
][0]
cmli = cml_interface(self.Base, parameters, iblock, task,
function, host)
cmli.run()
elif host == 'tensorflow':
# check methods
legal_functions = [
klass[0] for klass in inspect.getmembers(tfw)
]
if function not in legal_functions:
msg = "@function #%i: couldn't find function '%s' in the module '%s' wrarpper" % (
iblock, function, host)
raise NameError(msg)
self.references(host, function) # check references
self.Base.graph_info[iblock] = (host, function)
cml_interface = [
klass[1] for klass in inspect.getmembers(tfw)
if klass[0] == function
][0]
cmli = cml_interface(self.Base, parameters, iblock, task,
function, host)
cmli.run()
end_time = tot_exec_time_str(start_time)
tmp_str = "| ... done!"
print(tmp_str)
tmp_str = '| ' + end_time
print(tmp_str)
tmp_str = "=======\n\n"
print(tmp_str)
self._save_references()
tmp_str = "Total " + tot_exec_time_str(self.Base.start_time)
print(tmp_str)
tmp_str = std_datetime_str() + '\n'
print(tmp_str)